Pravah AI

Pravah AI is launching Indus-wx, a weather forecasting model built specifically for operators of India's electric grid. The product is aimed at renewable developers, utilities, and grid operators who need higher-resolution predictions than what global systems like ECMWF's IFS, its AIFS model, and NOAA's GFS provide for the subcontinent. The launch matters now because India's grid settles on 15-minute intervals while most public forecasts refresh only a few times a day, a mismatch that widens as solar and wind capacity grow and stress on the network compounds each year. Indus-wx issues a fresh forecast every hour, extends 48 hours out at 3 km resolution, and offers additional horizons at 10 days and 7 months. It fuses live satellite feeds, ground station readings, and numerical weather outputs, and is trained on Indian weather data to handle regional failure modes that global models miss, such as afternoon cloud behavior over the Deccan and low wind speed regimes in western Rajasthan. Pravah says the model has been benchmarked over a full year against leading global forecasts on grid-relevant variables, and the team is already piloting with renewable developers in India. The company was founded on the Stanford campus by Mohak Mangal, Dhruv Suri, Aman Gupta, and Nick Brown , and is part of the PearX S25 cohort building what it describes as foundation models for the electric grid. Indus-wx is the first public model in that broader effort, with the weather product available at weather.pravah.com.